Nanotribological behavior of ZnO films prepared by atomiclayer deposition

Wun-Kai Wang a,1, Hua-Chiang Wen b,1, Chun-Hu Cheng c, Wu-Ching Chou b,Wei-Hung Yau d,n, Ching-Hua Hung a, Chang-Pin Chou a

a Department of Mechanical Engineering, National Chiao Tung University, Hsinchu 300, Taiwan, ROCb Department of Electrophysics, National Chiao Tung University, Hsinchu 300, Taiwan, ROCc Department of Mechantronic Technology, National Normal University, Taipei 106, Taiwan, ROCd Department of Mechanical Engineering, Chin-Yi University of Technology, Taichung 400, Taiwan, ROC

a r t i c l e i n f o

Article history:Received 20 February 2013Received in revised form14 August 2013Accepted 18 September 2013Available online 12 October 2013

Keywords:A. Thin filmsB. Vapor depositionC. Mechanical properties

a b s t r a c t

We used atomic layer deposition to form ZnO thin-film coatings on Si substrates and then evaluate theeffect of pile-up using the nanoscratch technique under a ramped mode. The wear volume decreasedwith increasing annealing temperature from room temperature to 400 1C for a given load. Elastic-to-plastic deformation occurred during sliding scratch processing between the groove and film for loadingpenetration of 30 nm. The onset of non-elastic behavior and greater contact pressure were evident forloading penetration of 150 nm; thus, full plastic deformation occurred as a result of a substrate effect. Wesuspect that elastic–plastic failure events were related to edge bulging between the groove and film, withelastic–plastic deformation attributable to adhesion discontinuities and/or cohesion failure of theZnO films.

& 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Zinc oxide (ZnO), a metal oxide that form transparent films, isa wide-bandgap semiconductor with high thermal and chemicalstability and large exciton binding energy (�60 meV); it is one ofthe main materials used in blue and UV optical devices [1–3].ZnO can be deposited as a thin film using several techniques,including pulsed laser deposition, spray pyrolysis, and aqueoussolution methods [4–6]. In recent years, atomic layer deposition(ALD) has become an attractive method for the formation ofsmooth films, with the additional benefits of high conformalityand precise control over thickness and composition. ALD forma-tion of ZnO through cyclic self-limiting deposition can yieldconformal and uniform thin films at the atomic level [7,8]. Theelectrical, optical, and structural properties of such ZnO materialshave been investigated in detail for various processing para-meters and conditions [9].

ZnO films have recently been used in acoustic wave biosensingand microfluidic applications; their potential utility in the fabrica-tion of lap-on-chip (LOC) systems may affect the development ofthe biotechnology industry. Kim et al. reported that a passivated

ALD aluminum oxide surface could be used in an acoustic wavesensor for early biofilm detection [10]. Most acoustic wave devicesneed to be sensitive to input signals, such as mechanical stress,chemical exchange, and electrical perturbations. Thus, a strongtexture, a low density of defect traps, a smooth surface, and goodstoichiometry are basic requirements for good piezoelectric prop-erties. As mentioned above, ALD with precise compositionalcontrol using layer-by-layer deposition can meet these require-ments at a low thermal budget of 200 1C; the ability to integrateALD with CMOS processing is useful in the preparation of LOCsystems. However, the acoustic velocity of a ZnO film stronglydepends on substrate effects due to lattice mismatch, which limitsthe piezoelectric properties. The crystallinity of a ZnO film isresponsible for its piezoelectric response. Although the crystal-linity can be improved via high-temperature annealing, the inter-face quality will suffer from adhesion issues as a result of localstrain or stress originating from de-bonding of Zn atoms [11].

Nanoindentation can provide a good understanding of themechanical properties of a thin film interface, primarily itselastic–plastic properties, friction, and adhesion. These parameterscharacterize the mechanical stability of a working device and canprovide useful predictions regarding durability. Nanoindentation isperformed using an instrument that measures force and displace-ment continuously during indentation [13–18]. Slip band move-ment and dislocation nucleation mechanisms have been proposedto explain pop-in events within the epilayers of ZnO films [14–17].

Contents lists available at ScienceDirect

journal homepage: www.elsevier.com/locate/jpcs

Journal of Physics and Chemistry of Solids

0022-3697/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.http://dx.doi.org/10.1016/j.jpcs.2013.09.016

n Corresponding author. Tel.: þ886 423924505; fax: þ886 423930681.E-mail address: ywhung1@gmail.com (W.-H. Yau).1 These authors contributed equally to this work.

Journal of Physics and Chemistry of Solids 75 (2014) 334–338

Annealing treatment is commonly applied to minimize defects andimprove the quality of as-grown thin films. Kim et al. found thatannealing significantly improved the optical properties and sur-face morphology of ALD ZnO thin films compared to as-grownsamples [12].

In this study we investigated the nanotribological behavior ofALD ZnO films and the temperature-dependent crystallizationeffects of different annealing conditions. To gain an insight intothe failure mechanisms involved in nanotribological responses, weanalyzed the surface morphology, film crystallinity, microstruc-ture, and adhesion and cohesion using atomic force microscopy(AFM), X-ray diffraction (XRD), transmission electron microscopy(TEM), and scanning probe microscopy (SPM). We also examinedthe quality of the resulting ZnO layers in terms of pile-up eventsunder sliding scratch-induced deformation.

2. Experimental

Water and diethylzinc were used as precursors and nitrogen asthe purge gas. Thin ZnO films were deposited on Si (1 0 0)substrates at 200 1C in an ALD system operated using the flow-rate interruption method [8,19]. The number of deposition cycleswas fixed at 150 in each experiment and the deposition rate wascontrolled at 0.2 Å per cycle. The thickness of each as-depositedZnO film was approximately 30 nm. As-deposited ZnO films weresubjected to rapid thermal annealing at 300 or 400 1C for 1 hunder N2. The heating rate was set at 20 1C s�1; the furnacerequired approximately 30 min to cool to RT after annealing. TheZnO crystallinity was analyzed using XRD (PANalytical X'Pert Pro,MRD) with Cu Kα (λ¼0.154 nm) radiation. Scans were conductedover the 2θ range 20–801 (scan rate 21 min�1) at a grazing angle of0.51 under 30 kV and 30 mA. XRD data conformed to the JointCommittee on Powder Diffraction Standards card (JPCDS). Cross-sectional TEM (JEOL JEM-2010F, operating voltage 200 kV) wasused to examine the ZnO/Si microstructure.

The nanotribological properties of ZnO films were determinedusing an SPM system in which AFM (Digital Instruments Nano-scope III) was combined with a nanoscratch measurement system.The scanning probe microscope (Hysitron Corporation) was oper-ated at a constant scan rate of 2 μm s�1. The adhesion propertiesof as-deposited samples and samples annealed at 300 and 400 1Cwere evaluated using a nanoindentation system. The constantdepth mode was used and penetration depths of 30 and 150 nmwere applied for each ZnO film. The maximum load was main-tained while forming 10-μm-long scratches. Surface profilesbefore and after scratching were obtained by scanning the tip ata normal load of 0.02 mN (i.e., a load sufficiently small to produceany measurable displacement). After scratching, the wear trackswere imaged using AFM.

3. Results and discussion

Fig. 1 shows XRD patterns for as-deposited ZnO and samplesannealed at 300 and 400 1C. The diffraction peaks for ZnO (1 0 0),(0 0 2), (1 0 1), (1 0 2), (1 1 0), (1 0 3), and (1 1 2) evident in theXRD patterns confirm that all the films were polycrystalline with ahexagonal wurtzite structure. The intensity of the ZnO (0 0 2) peakwas higher for the sample annealed at 400 1C than for the filmannealed at 300 1C, indicating that the higher annealing tempera-ture increased the crystallinity. The results suggest that the ZnOlayer remained largely polycrystalline with excellent crystallinequality.

We obtained more detailed structural information using TEM.Fig. 2 shows a cross-sectional TEM image of the ZnO/Si system,

which features a layer of polycrystalline ZnO. Randomly orientedZnO nanocrystals within the 30-nm-thick ZnO layers appear tohave very good crystallinity. A nanometer-thick interface region isevident between Si and the ZnO layer. This interface appearsblurred and the ZnO grains it contains have relatively poorcrystallinity. As described below, this interface has a significanteffect on the ZnO/Si adhesion properties.

Fig. 3 presents typical AFM profiles of ZnO films deposited on Sibefore and after nanoscratch tests at a penetration of 30 or150 nm. The scratch images reveal that an increase in load led toan increase in wear volume (total volume swiped by the indentertip) for each ZnO coating; the wear volume is consistent with theprojected area of the indenter tip. Using such a scratch system todetermine elastic–plastic contact, Pelletier et al. found that theshape of the residual groove was related to the plastic strain field[20]. The profile in Fig. 3a indicates that the surface materialunderwent an elastic reaction as a result of elastic deformationbetween the groove and the film when the penetration depth was30 nm. The indenter resulted in greater contact pressure for thesample annealed at 300 1C, leading to the onset of non-elasticbehavior (Fig. 3b). For the sample annealed at 400 1C, large areas ofthe coating were removed from the scratch track as a result ofelastic–plastic deformation at 30 nm (Fig. 3c). The penetrationdepth traces reveal a slight increase for the sample treated at400 1C compared to the sample treated at 300 1C. This SPM trendmatches the lateral force curves in Fig. 4a, from which it is evidentthat an elastic reaction between the groove and film dominates.A similar response is evident in Fig. 3d–f: an increase in lateral

Fig. 1. XRD patterns for ALD ZnO films. (a) As-deposited film and films annealed at(b) 300 1C and (c) 400 1C.

Fig. 2. Cross-sectional TEM image of the ZnO/Si system.

W.-K. Wang et al. / Journal of Physics and Chemistry of Solids 75 (2014) 334–338 335

force is evident for a penetration depth of 150 nm, withoutmarked oscillations. However, the substrate effect may dominateas a result of the increased lateral force.

The annealing temperature was sufficient to drive transforma-tion to a denser crystal structure so that ductile deformation wasinhibited. The critical load was 29.8 and 50.6 μN for the samplesannealed at 300 and 400 1C, respectively. No signal was detectedfor the as-deposited sample. Sample annealing at 400 1C led torelaxed crystallization of the ZnO coating, as observed in Fig. 4a,b.The changes in scratch groove morphology also indicate animprovement in the nanoscale wear resistance. At the beginningof each scan, all of the films exhibited similar behavior, but theeffect of the annealing temperature became apparent thereafter.This behavior may be due in part to shrinkage events, attributableto a very thin coating that responds to a susceptible defect densityfor scratch tests at a penetration depth of 30 nm. Schreiber andVasnyov used dynamic-mode cathodoluminescence to observelocal plastic deformation during in situ scratching of a (0001)ZnO sample [22]. They found evidence of the propagation of highly

mobile dislocations in various prismatic slip systems for theiras-deposited samples. Thus, we suspect that the features of thenanoscratch zone in the vicinity of the interface might not bemerely accidental artifacts resulting from sample preparation [18].Additional evidence will be required because this phenomenoncontrols oscillations under friction, thereby affecting the adhesionand/or cohesion failure of ZnO films. Fig. 5a shows the frictioncoefficient recorded at a penetration depth of 30 nm. Oscillationsunder the friction force are evident, particularly for an annealingtemperature of 400 1C. Owing to thin film elastic deformation, thefriction coefficient decreased, gradually approaching the truevalue; elastic–plastic deformation of the friction coefficient wasslightly disrupted in the interval 15–25 s. At 22.5 s, film dislocationcaused oscillation of the friction coefficient, which then stabilizedat 25 s. The friction coefficient recorded at a penetration depth of150 nm (Fig. 5b) also exhibited oscillation under the friction force.Similar analysis of scratched films has been reported previously[19,20]. Notably, oscillation trends for each sample were similar inboth on- and off-load scans [21]. The oscillations in the plots

Fig. 3. AFM images combined with nanoscratch trace measurements for nanoscratched samples obtained using a ramp force at a depth of (a–c) 30 nm and (d–f) 150 nm.(a,d) As-deposited sample; (b,e) sample annealed at 300 1C; and (c,f) sample annealed at 400 1C.

W.-K. Wang et al. / Journal of Physics and Chemistry of Solids 75 (2014) 334–338336

recorded at a penetration depth of 150 nm are more stable thanthose recorded at 30 nm. We suspect that the lower degree ofadhesion reflects interlinks and rearrangements that tend to resultin fluctuations of the friction coefficient (μ) at a penetration depthof 150 nm. Instability in the initial μ profile, so-called pile-up, isevident in Fig. 5a for the sample annealed at 300 1C. We calculatedvalues of μ for the ZnO films by averaging results measured atscratch durations ranging from 8 to 38 s (Fig. 6). From the trendsobserved in our nanotribological investigations, changes in thevalue of μ appear to signal the onset of adhesive failure, withcracking or delamination resulting between the interface and thesliding nanoscratch track [15,23].

The discordant curves and irregularities observed during elasticand plastic deformation can be attributed to adhesion disconti-nuities and/or cohesion failure, as determined from the nano-scratch traces (induction of pile-up) [24,25]. The profile for thesample annealed at 400 1C was softer than for the as-depositedsample owing to shrinkage of the nanoscratch traces. Thus, theZnO film structure changed when the thermal treatment tem-perature was increased from 300 to 400 1C, presumably viaalteration of the composition of the intrinsic structure. Hence,edge bulging occurred between the groove and film and thematerial was crushed as a result of elastic failure. In summary,the friction traces were fairly reproducible, but varied greatly forthe different samples.

4. Conclusion

We used AFM and nanoscratch techniques to characterize high-quality ALD ZnO films on Si substrates. We performed nanoscratchtests to determine the adhesion/cohesion failure mechanism of

Fig. 4. Normal displacement plotted with respect to scratch duration for ZnO filmsannealed at 300 and 400 1C. Each sample was subjected to penetration at (a) 30 nmand (b) 150 nm.

Fig. 5. Coefficient of friction (μ) plotted with respect to scratch duration foras-deposited ZnO film and films annealed at 300 and 400 1C for penetration depthsof (a) 30 nm and (b) 150 nm. Inset to (a): 15–25-s interval.

Fig. 6. Distinct critical coefficient of friction for ZnO films annealed at varioustemperatures and then subjected to penetration at depths of 30 and 150 nm.

W.-K. Wang et al. / Journal of Physics and Chemistry of Solids 75 (2014) 334–338 337

ZnO-coated Si. An increase in annealing temperature was suffi-cient to drive transformation to a denser crystal structure so thatductile deformation was inhibited. Pile-up occurred at a penetra-tion force of 29.8 and 50.6 μN for samples annealed at 300 and400 1C, respectively. The systems exhibited significant suppressionof lateral indented deformation, which induced a degree of pile-up. In addition, tip drop and lift phenomena can be explained interms of film stiffness as a function of interatomic distances duringscratch tests. For ZnO films compressed by an indenter, more μoscillations were observed for a penetration depth of 30 nm thanfor 150 nm. It is reasonable to suspect that edge bulging occurredbetween the groove and film, resulting in elastic failure.

Acknowledgment

This study was supported by the National Science Council ofTaiwan (grant nos. NSC 101–2221-E-167-004 and NSC 102–2221-E-167-007) and the National Chin-Yi University of Technology(NCUT11-REM-004: 2012.1.1–2012.10.31 and NCUT 12-REM-003:2013.1.1–2013.10.31). We thank Dr. H.C. Wen and Prof. W.C. Choufor technical suggestions, and Prof. Y.R. Jeng for the SPM equip-ment used in this study.

References

[1] A.P. Alivisatos, Science 271 (1996) 933–937.[2] K.C. Aw, Z. Tsakadze, A. Lohani, S. Mhaisalkar, Scr. Mater 60 (2009) 48–51.[3] Y.R. Ryu, T.S. Lee, J.A. Lubguban, H.W. White, B.J. Kim, Y.S. Park, C.J. Youn, Appl.

Phys. Lett. 88 (2006) 241108.

[4] S. Christoulakis, M. Suchea, E. Koudoumas, M. Katharakis, N. Katsarakis,G. Kiriakidis, Appl. Surf. Sci. 252 (2006) 5351–5354.

[5] I. Kortidis, K. Moschovis, F.A. Mahmoud, G. Kiriakidis, Thin Solid Films 518(2009) 1208–1213.

[6] X.X. Liu, Z.G. Jin, S.J. Bu, J. Zhao, Z.F. Liu, J. Am. Ceram. Soc. 89 (2006)1226–1231.

[7] Y.H. Lin, Y.C. Lai, C.L. Lu, W.K. Hsu, J. Mater. Chem. 21 (2011) 12485–12488.[8] H.C. Wen, C.I. Hung, H.J. Tsai, C.K. Lu, Y.C. Lai, W.K. Hsu, J. Mater. Chem. 22

(2012) 13747–13750.[9] Y.W. Zhu, H.I. Elim, Y.L. Foo, T. Yu, Y.J. Liu, W. Ji, J.Y. Lee, Z.X. Shen, A.T.S. Wee,

J.T.L. Thong, C.H. Sow, Adv. Mater. 18 (2006) 587–592.[10] Y.W. Kim, S.E. Sardari, M.T. Meyer, A.A. Iliadis, H.C. Wu, W.E. Bentley,

R. Ghodssi, Sens. Actuators B 163 (2012) 136–145.[11] T. Riemann, J. Christen, G. Kaczmarczyk, A. Kaschner, A. Hoffmann, A. Zeuner,

D. Hofmann, B.K. Meyer, Phys. Status Solidi b 229 (2002) 891–895.[12] C.R. Kim, C.M. Shin, J.Y. Lee, J.H. Heo, T.M. Lee, J.H. Park, H. Ryu, C.S. Son,

J.H. Chang, Curr. Appl. Phys. 10 (2010) S294–S297.[13] C.Y. Yen, S.R. Jian, G.J. Chen, C.M. Lin, H.Y. Lee, W.C. Ke, Y.Y. Liao, P.F. Yang,

C.T. Wang, Y.S. Lai, J.S.C. Jang, J.Y. Juang, Appl. Surf. Sci. 257 (2011) 7900–7905.[14] W.H. Yau, P.C. Tseng, H.C. Wen, C.H. Tsai, W.C. Chou, Microelectron. Reliab. 51

(2011) 931–935.[15] M.H. Lin, H.C. Wen, Y.R. Jeng, C.P. Chou, Nanoscale Res. Lett. 5 (2010)

1812–1816.[16] M.H. Lin, H.C. Wen, C.Y. Huang, Y.R. Jeng, W.H. Yau, W.F. Wu, C.P. Chou, Appl.

Surf. Sci. 256 (2010) 3464–3467.[17] H.C. Wen, C.S. Yang, W.C. Chou, Appl. Surf. Sci. 256 (2010) 2128–2131.[18] S.R. Jian, I.J. Teng, P.F. Yang, Y.S. Lai, J.M. Lu, J.G. Chang, S.P. Ju, Nanoscale Res.

Lett. 3 (2008) 186–193.[19] C.J. Hu, Y.H. Lin, C.W. Tang, M.Y. Tsai, W.K. Hsu, H.F. Kuo, Adv. Mater. 23 (2011)

2941–2945.[20] H. Pelletier, A.L. Durier, C. Gauthier, R. Schirrer, Tribol. Int. 41 (2008) 975–984.[21] J. Zhang, J. Zhang, Tribol. Lett. 49 (2013) 77–83.[22] J. Schreiber, S. Vasnyov, J. Phys. Condens. Matter 16 (2004) S75–S84.[23] R. Saha, W.D. Nix, Acta Mater. 50 (2002) 23–38.[24] Y.R. Jeng, P.C. Tsai, Y.H. Liu, Mater. Res. Bull. 44 (2009) 1995–1999.[25] K. Shinozaki, T. Honma, T. Komatsu, Mater. Res. Bull. 46 (2011) 922–928.

W.-K. Wang et al. / Journal of Physics and Chemistry of Solids 75 (2014) 334–338338